Study on Natural Language Generation for Spatial Information Representation

In this paper, based on spatial cognitive theories connected with the construction of natural language generation model for spatial information representation, we research on related spatial orientation results from the study field of Modern Chinese, analyzed semantic mapping between spatial information represented by kinds of maps and spatial orientation information expressed by natural language, and bring spatial concept into the background of linguistics and GIS from a unique perspective. We determine the spatial information that should be represented in this kind of natural language, as well as set up intermediate semantic model which can be described by spatial point of point. This study is a contribution to summarize the macroscopic research achievements including spatial conception related to natural language, and generalize the microcosmic methods with respect to object-based spatial model and natural language generation. Implementation of a semantic framework mapped between spatial information and natural language is important from a methodological point of view because it guides us to build spatial knowledge base which formalize the semantics of natural language and speck spatial relation in a computational model which allows for similarity comparisons. This paper describes an experiment that investigates text generation result for bus route-finding.